0

RWD basert på register- och journaldata

Mats Rosenlund, PhD, MPH

Principal, Head of Real World Evidence Solutions

and HEOR, Nordics, IMS Health, Stockholm

Center for Pharmocoepidemiology, unit of Clinical Epidemiology,

Department of Medicine, Karolinska Institutet

1

Agenda outline - 4 questions covered

Why is the interest for RWD data increasing?

What is driving the need for RWD? What kind of RWD can be generated from register and EMR?

How is RWD used across the product life cycle by payers, regulators, and industry?

When is RWD important?

2

Patient data reflects a seismic shift in evidence evaluation

Few evaluators at launch, mostly

regulators and large payers

Many groups over time

including clinical and

small payers

THE PAST

THE PRESENT

RCT

Few

Many

Efficacy and Safety Almost everything

Initial view of

benefit-risk

Insights on environment,

outcomes, costs,

comparative effectiveness

Controlled trials,

manufacturer led

Shift to secondary patient-level

data across sources

RCT and RWE

3 IMS Health Confidential

Drivers of change

• Increasing cost pressures

• New treatments to fund

• Providers evaluated on quality

• Increase in post-authorization

regulatory requirements

• Adaptive licencing

• Need for differential value proposition

• Need for R&D, clinical trial efficiency

• Explosion in volume of electronic

patient data

• Ability to bring data across silos

• New technologies and

methodologies

• New stakeholders conducting

analysis

Increasing patient-level data

and analytic technologies More scrutiny by

more stakeholders

SUPPLYY DEMAND

RWD & RWE

4

Supply: Growth in Real-World Data (RWD)

“90% of the world's data has been produced in just the last two years”

IMS Health Confidential

RWD is PATIENT data

Pharma data

(observational)

Electronic medical

and health records

Social

media data

Consumer

data

Claims

databases

Hospital

data

Disease

registries

Mortality

data

Pharmacy

data

Lab/Biomarkers

data

- U.S. Chamber of Commerce Foundation

5

Importance of different data types

Different perspectives among payers and pharma industry

Payers rate comparative

data and RWD higher than

pharmaceutical companies,

while RCTs are percieved as

less important

6

RWE intensifying across the product lifecycle, particularly around launch

Example below is RWE curve for a key oral anti-diabetic launch

Past

Now

Evid

en

ce

re

qu

ire

d

Development Growth phase Mature phase

Launch

Understand standard of care

Trial design

Patient recruitment

Unmet need / disease burden

Budget impact

Post marketing commitments (safety etc.)

Conditional pricing review

Utilization /prescribing patterns

Adherence

New competition

New formulation/ indication

Competitor goes generic

Long-term clinical outcomes

Differentiation in sub-populations

Head to head comparative effectiveness

Target populations

UsageDiffer-ence

Effects of switching on outcomes

Differentiate with or vs. protected galenics

Input to C/E models, e.g. comparator, dosing, discontin.

Fore-casting

Price-volume agreements

Maturity for RWE and Value Based Health CareSource: BCG report; Progress toward value-based health care, 2012

8

Nordics have outstanding opportunities in Real World Evidence

● Well structured public

health care system

● Universal implementation

of EMRs since 1990’s

● National and population-

based health registers in

the region since 1950’s

● Common set-up of health

care, health information

and classification systems

● Strong academic hubs to

collaborate with in the

region

99

RWD from Nordic registries

Electronic

Medical

Records

Socio-

economic

registry

Patient

registry

Cause of

death

registry

Prescription

registry

Primary

care

databases

Disease

specific

registry

Via CXP

Tailored study dataset

capturing full patient journey

10

Application for extracting relevant data from EMRs

Modern extraction tool that delivers anonymized patient data for clinical studies

Supports extraction from different health care (EMR) systems at hospitals, GPs,

psychiatric care, geriatrics, spirometry, E-archive systems, etc

Makes the complex task of collecting data from different sources into a single,

standardized and quick task

International development focus

Pygargus CXP 3.0

11

What data can we extract from EMRs?

Basically all patient data added into the EMR

Demographic data (age, gender etc.)

Prescriptions (ATC code, brand name, dose, pack size, strength etc)

Drugs administrated in

hospitals

Diagnoses/co-morbidities (ICD 9

or 10)

All lab tests performed (date,

test value)

Type of HCP contacts (doctor, nurse, other HCP

etc)

Patient data (weight, length,

BMI etc.)

Pathology

Surgical reports

Measures (Blood pressure, BMI, …)

Free text, search by stick words

12 IMS Health Confidential

Emerging trend: focus on creating T-Shape Evidence Networks that balance data source breadth and depth

T-Shaped Evidence Network: A set of relationships, spanning different organisations,

which provides access when required to high quality, clinically rich data

Broad data: RWD

sources spanning

broad, near population

level cohorts with

minimal clinical

information about

each patient

Deep data: Clinically

rich data for a discrete

population

of patients typically

with coverage of at

least 5% of a total

market population1

Nationally relevant databases

Augmented datasets

Biobanks, labs, registries

BROAD

DEEP

Ran

ge

of d

ata

typ

es

Deep, flexible

Clinically rich symptom / outcome data in Epilepsy or

Parkinson Disease

To create a complete view, we

use sophisticated approaches

to supplement data including:

eCRF, ePRO, NLP and rapid

custom sourcing/linkage

1 Estimation based on incidence rates

13

T-shaped data strategy to increase study power, external validity, and clinical depth

BROAD

• Demographics

• Co-morbidities

• Survival

• Diagnoses

• Concomitant

medication

• Procedures

• Prognostic factors

• In-patient treatments

• Disease progression

• Treatment response

• Laboratory test results

Validate

representa-

tiveness of the

EMR cohort

Patient

demographics

Prescribed drugs

Co-morbidities

Mortality rates

DE

EP

The ”big” that really matters in big data is not breadth – not more patients, but rather more variables covering the entirety of the patient experience.

[E&Y. Global Pharmaceutical Report 2014]

14

The process from protocol to dissemination

Design protocol

EC approval & site

recruitment

Data extraction with

CXP

Data mapping & cleaning

Data linkage and

transformation

Data management

Analysis

Presentation of results

Publication

Preparation and extraction

Data management

Analysis and dissemination

Extracted/raw database

DB DB

Study database

… based on > 40 projects

Hospital EMRs

Primary care EMRs

Timelines vary according to complexity

15

Example of RWD

• To better understand the epidemiology of COPD including treatment practice, co-morbidity and mortality during the last decade

Overall rationale

• Describe the long-term epidemiology of COPD and the health care structure in primary care

• Assess the comparative effectiveness between fixed ICS/LABA combination treatment in COPD

• To investigate occurrence of pneumonia in a COPD population treated with fixed combination

• Estimate the cost-effectiveness amongst fixed combination treatments in COPD

Study Objectives

16

COPD Case study

Background • COPD associated with considerable morbidity, mortality;

pneumonia is a frequent complication

• To prevent exacerbations, patients are frequently prescribed fixed-dose ICS/LABA

combinations

• Pharma client wanted to develop credible insights about the best treatment course for

COPD patients to work with stakeholders to improve guidelines, COPD management

The Study Retrospective, observational,

population-based study of 21,361

COPD patients in Sweden over 11 years

to focus on

• Prevalence and incidence

• Co-morbidities

• Disease management

• Complete clinical data on each patient

• Mortality and life expectancy

The results• Improved outcomes

− Earlier diagnosis

− Primary care focus

− Changed treatment options

− Decreased exacerbations

• TA-specific impact on exacerbations,

pneumonia-related mortality

• 4 publications in BMJ and J Int

Medicine, Prim Care Resp Journal

• 5 scientific abstracts presented

1. Jansson C, et al. BMJ 2013; 346:1-11; 2. Larsson K, et al. J Intern Med 2013; 273 (6): 584-94

17

The study designRetrospective, observational, population-based, propensity score matched study

• EMR data from 76 primary health care centres linked with

data from Swedish registries (8% of the Swedish pop)

• Patients followed from January 1999 to December 2009, or

end of treatment, emigration or death

The PATHOS study in numbers

27 934 Patients with COPD or treatment

21 361 COPD patients

190 000 Visits at hospital

3 500 000 Contacts with primary care

1 300 000 Prescriptions

18

• Pairwise matching with 2734 patients per

treatment group including 19,170 PY’s

follow-up with either ICS/LABA

combination

• Only 4 patient from the smaller treatment

group could not be matched to a

corresponding patient within the database

• Data 2 years before index date used for

matching

Propensity score matching of populations at index to limit potential for bias

* It was not possible to match for patient weight, height, BMI and smoking, as these

measurements were available for just a minority sub-sample.

Propensity score matching was performed to account for individuals disease severity*

Number of eligible patients by treatment

2738 Patients using Seretide

7155 Patients using Pulmicort

9893 Total number of COPD patients

Variables included in propensity score

Demographics Age, gender, time from COPD

diagnosis

Medication Antibiotics, SABA, oral/inhaled

steroids, LABA, anticholinergics,

CV medications

Hospitalisation For exacerbations, any

cardiovascular reason,

pneumonia and asthma

Co-morbidities Diabetes, asthma, cancer, heart

failure, hypertension, stroke,

FEV1 % pn

19

27%

26%

-1%

32%

37%

Difference

0,36

0,18

0,82

0,38

0,71

0,58

0,26

0,81

0,52

0,98

0,00 0,20 0,40 0,60 0,80 1,00 1,20

Number of hospitalizations

Emergency room visits

Oral steroids

Antibiotics

Exacerbations

Yearly rateCenters without asthma/COPD nurse

Centers with asthma/COPD nurse

Stratification by primary care centers with and without asthma/COPD nurse

Rate of exacerbations and hospitalizations lower in PCC with COPD nurse

20

The study demonstrated better disease management, safety, and comparative effectiveness

Comparative effectiveness between two treatments in

preventing exacerbations in COPD

Disease

management

evolution over

11 years

showed

dramatic

decrease in

number of

exacerbations 1,3

4,2

7,4

11,0

0,7

2,7

4,3

6,4

0 2 4 6 8 10 12

Hospital outpatient …

Primary care diagnosis

Pneumonia hospitalizatio…

Pnuemonia diagnoses

BUD/FORMSAL/FLU

Event rate per 100 patient–years

Pneumonia and related mortality in COPD

treated with fixed combinations of inhaled

corticosteroid and long acting β2 agonist

Safety

21

Publications from the study